A model for Distributed Machine Learning

Nowadays, distributed machine learning has become essential, given the limitations of traditional centralized approaches that rely on central model training and data collection. Although distributed AI has proven efficient and adaptable in overcoming these constraints, it still faces important challenges, especially in dynamic, heterogeneous, and resource-limited environments. To address these issues, this paper discusses the main approaches and current limitations of distributed machine learning within the Cloud–Edge–IoT continuum (CEI) and highlights our ongoing work on integrating intelligent monitoring mechanisms as a necessary enhancement. This type of monitoring makes distributed learning more adaptive, efficient, and stable.